The fingerprinting technique based on received signal strength (RSS) is widely used for indoor positioning. However, RSS-based indoor positioning has a problem because of the RSS variation due to the structure of the indoor environments and cross-device mobile station. This study presents the DIFF-based indoor positioning system using fingerprinting technique and K-Means clustering based on measurement data. It was found that the DIFF-based fingerprinting technique with no-clustering case was provided slightly higher accuracy than K-Means clustering case. However, the K-Means clustering case computed fast than no-clustering case many times. Therefore the K-Means clustering case can be suitably used for the applications on the mobile devices that have limited memory and require fast computation.